基于诱发电位特征分析的脊髓损伤模式及位置精准诊断方法研究

Multimodality intraoperative monitoring with somatosensory and motor evoked potentials has been widely used in early warning and protection of spinal cord injury. Howver, it can only reflect the intergrity of spinal cord function, without precise detection of the mode and location of injury. Previose experimental studies reported that different iatrogenic spinal cord injury model (such as compression, stretch, dislocation) can lead to injury on different structures in the cord, resulting in fifferent patterns of electrophysiological signals. Several recent studies further demonstrated that the components of evoked potential could indicate the location of the spinal cord injury. This project will establish various animal models to simulate iatrogenic spinal cord injuries of different damage modes at different locations. The component analysis of multi-channel somatosensory and motor evoked potentials will be investigated by using machine learning and classification to big data from a large scale animal experiments with different injury modes and locations. This study will explore the new features of evoked potentials, which may provide a novel method to precise detection of mode and location of spinal cord injury. The results from this study may be trasnlated to clinical application in a totally new intraoperative monitoring technology for spinal surgery. It will establish a solid scientific base for prevention, intervention and prognosis of iatrogenic spinal cord injury during spinal surgery.

体感和运动诱发电位联合监测被广泛使用于术中脊髓损伤的预警与防护，但只能整体反映脊髓神经功能的状况，不能对脊髓损伤模式和位置进行精确判断。前期实验研究发现，不同医源性脊髓损伤模式（如压迫、牵拉、错位等）会造成不同脊髓结构的损伤，其组织学和电生理均有不同的表征；最近的一些研究还发现了诱发电位的成份分析可以反映脊髓损伤的位置信息。本课题将建立脊柱不同位置、不同损伤模式的医源性脊髓损伤动物模型，对多通道体感和运动诱发电位进行信号时频谱成份分析，所获得的大量数据将利用机器学习分类的方法分析不同位置和损伤模式的诱发电位模式特征分类模型，探索可以特异性地监测脊髓损伤模式和位置的诱发电位检测新方法。研究成果可以转化为监测不同损伤模式和位置的医源性脊髓损伤新方法，将为脊柱手术中脊髓损伤的预防、干预及其术后治疗预案提供崭新的科学手段。

体感和运动诱发电位技术被广泛使用于术中脊髓损伤的临床监护。然而，诱发电位目前只能反映脊髓神经的整体功能，不能准确反映脊髓损伤的模式和位置。我们的研究发现，不同医源性损伤模式（如脊髓的压迫、牵拉、错位等）会造成不同脊髓结构的损伤，其组织学和电生理均有不同的表征。通过对诱发电位成份分析进行特征识别，对大数据进行机器学习分类，可以建立对不同位置和损伤模式的诱发电位模式特征分类模型。本项研究已经在实验研究中证实了这一科学假设，我们还研究了动物与人类电生理信号特征的相关性，所获得的研究成果可以进行临床转化，成为监测不同损伤模式和位置的医源性脊髓损伤新方法，将为脊柱手术中脊髓损伤的预防、干预及其术后治疗预案提供崭新的临床手段。